Hyperconverged IT infrastructures: theory and practice

vom 20.01.2021

30.12.2020 I by Henrik Hasenkamp

Hyperconverged systems provide the entire data center as a virtualized service.In addition to many advantages, the concept also has a few stumbling blocks to offer, which is why companies should carefully consider them before implementing them.

The better IT systems are coordinated, the less error-prone they are and the easier it is to maintain them. In practice, this so-called convergence emerges from the experience of the IT team over many years. The goals are clear: to meet the growing demands on IT while simplifying the infrastructure as much as possible. Individual components are often combined to form an appliance - preconfigured, quickly deployable or bookable as a service.Technologies such as virtualization and cloud computing have given the development of convergence an enormous boost in recent years. Adding more appliances ensures scalability, and especially when cloud services are used, IT remains flexible and demand-driven. Nevertheless, the structure of convergent infrastructures remains classic: individual proprietary systems consisting of hardware components with the associated software. The concept of hyperconvergence (Hyperconverged Infrastructure, HCI) is the consequence of this: by virtualizing the entire IT infrastructure, a comprehensive appliance is created, so to speak. The software of the individual components is replaced by an overarching intelligence. For the user, the result is IT that is geared exclusively to his or her needs: The user's applications request resources via an abstract level with virtual servers, and the system allocates them. In addition to central administration, the more flexible utilization of IT resources is one of the main arguments in favor of hyperconvergence.

Hyperconverged IT infrastructures for every company size?

The advantages of HCI technology make it interesting for any size of company - at least in principle. The pre-configured systems come as a well-tuned complete package, and within a very short time the customer has a data center available, either on-site or as a service. Companies trans-forming from a hardware-centric infrastructure to a virtualized, more flexible environment should look at HCI. The on-demand allocation of computing capacity has a positive impact on performance; therefore, infrastructures with virtual desktops can be implemented just as well with HCI as performance-intensive scenarios such as DevOps or IoT (Internet of Things).But while the complexity of the hardware is decreasing, that of the software is increasing. While the operating systems of the individual devices no longer need to be coordinated with each other, the virtualization software, the so-called supervisor, becomes the complex linchpin. This requires appropriate know-how, which is not directly available in every company. Switching to HCI also involves a paradigm shift, which is associated with costs and risks. Small companies in particular should always check the service offering during the conception phase in order to keep investment costs low. In addition, the implementation of HCI sometimes leads to oversizing, because HCI systems are only scalable as an overall appliance. The new HCI box is comparatively easy to add and connect. But if, for example, only additional storage is required, the costs quickly exceed the benefits. Another disadvantage, especially for very individually structured IT environments: HCI systems are prefabricated solutions - they do not always meet all the special requirements of a company.

The choice of hypervisor

HCI's rapid implementation speaks to the technology, but companies should still take the time to review multiple offerings in advance. The choice of vendor and its hypervisor is critical. Since the HC infrastructure is provided as a complete package by one vendor, the company places itself in a certain dependency. This is because the hypervisor is the central component for virtualization: it creates the virtual machines (VMs) and runs them. It allocates IT capacity to the VMs, which it manages in a resource pool. Most HCI offerings rely on standard hypervisors, such as VMware's vSphere and Microsoft's Hyper-V. In the open source space, KVM (Kernel-based Virtual Machine) has established itself as the de facto standard. These have an immense range of functions, are very complex and multi-layered, but are also widely accepted and compatible. Proprietary hypervisors are usually more specifically adapted to their tasks and are easier to use due to their smaller feature set. However, it can be more difficult to find suitable know-how, and at the same time this increases the dependency on the provider. Connection to other existing infrastructure components could also be more complicated than with a standard product.

HCI as a managed service

The acquisition of an HCI is a cost factor that should not be neglected; after all, you are also acquiring a fully functional data center. So it is an attractive alternative, especially for medium-sized companies, to buy the whole thing as a service. An HCI Managed Service is an infrastructure service from the cloud - with all the advantages that speak for cloud computing. The high investment costs are eliminated, and the cloud service provider takes over large parts of the IT management, including physical security and software-based security. For companies, this offers the opportunity to test HCI at low risk and to gain experience without immediately committing themselves to a fixed provider. Many companies are thus in a position to purchase higher-quality and higher-performance IT resources than they could provide themselves.The flexibility and scalability of the cloud help to keep an eye on costs and absorb any know-how bottlenecks.

From theory to practice: implementation

To weigh up whether a hyperconverged appliance is appropriate, companies should make some preliminary considerations. The specific need and the existing IT infrastructure are the two most important influencing criteria. Whether the solution can be integrated must also be clarified, as must questions of vendor support and the procurement of know-how. The hypervisor is of particular importance - which functions it should offer, how complex it is to operate and how integrative it is must be clarified in advance. If a managed service solution is to be implemented, corresponding agreements with the cloud provider are necessary.Last but not least, the decisive, overarching point is the costs. Companies should critically compare convergent solutions as a supplement to the existing infrastructure with HCI in their own data center and cloud offerings. Depending on the time period considered, the results will be very different. Finally, the implementation itself is comparatively straightforward. Although there is still some integration work, it is significantly less than with conventional IT installations.

The original article in german can be found here.

    Back to overview